Open-Notebook

Open Notebook is a privacy focused alternative to Google's Notebook LM. Research goes into a system you run yourself, so sensitive documents never leave your infrastructure and no single vendor decides which model gets to read them.

It is built with Python, Next.js, React, SurrealDB and LangChain, and ships as a Docker Compose stack with the database and the application as separate services. API keys are configured in the interface after startup rather than in files.

The feature set targets the parts of research that a closed notebook tends to restrict.

• Model choice: works with more than 18 providers including OpenAI, Anthropic, Ollama and LM Studio, so running locally is a real option.
• Multi-modal sources: organizes PDFs, videos, audio and web pages in one place.
• Podcast generation: produces episodes with one to four speakers and custom speaker profiles.
• Search: covers all stored content with both full text and vector queries.
• REST API: exposes the whole notebook for automation, which the Google product does not offer at all.

The interface is available in English, Portuguese, Chinese in simplified and traditional forms, Japanese, Russian and Bengali. Docker Desktop is the only prerequisite, the containers expose a web UI and a REST API on separate ports, and costs stay limited to whatever AI usage you choose to pay for.